Deep Learning for Accurate Detection of Brute Force attacks on IoT Networks
Ahmed Fawzi Otoom, Wafa’ Eleisah, Emad E. Abdallah
Abstract
Internet of Things (IoT) area is fast growing and wide spreading with its application being present to our daily lives. Communication between IoT devices is controlled using various types of protocols. A popular example of these protocols is Message Queue Telemetry Protocol (MQTT) which is a lightweight and reliable protocol for communication. However, there have been cyber-attacks that targeted MQTT-IoT networks which raises the attention for an efficient intrusion detection system for detecting such attacks. Popular type of such attacks is brute force attack. In this paper, we propose deep learning for automatic detection of brute force attacks on MQTT-IoT networks. We use a recent dataset, MQTT-IoT-IDS2020 dataset, to train the deep learning model with high number of instances and using flow-based feature. The classification model is very accurate in detection of such attacks with more than 99% accuracy in discriminating between normal and brute force attacks.